LECTRONIC DATA APTURE - · PDF fileAll these methods result in the data being available for further ... Connectivity - Can permanent ... disadvantages and indeed,

ELECTRONIC DATA

CAPTURE

Jonathan TilburyTESSELLA SUPPORT SERVICES PLC

Issue V1.R1.M0April 2003

Issue: V1.R1.M0 Copyright © Tessella 2003

ELECTRONIC DATA CAPTURE

What is Electronic Data Capture?Electronic Data Capture (EDC) is the gathering of data collected by humans intocomputer systems without the need for manual data re-entry. Although automaticdata entry has been around for a number of years – who can forget those pencil-based multiple-choice exams from years ago – the constant improvements in frontend technology have meant that the functional richness of the process is growingand the benefits are becoming more pronounced.

Data entry can be achieved using a number of mechanisms. Users can enter datadirectly into an electronic device such as a laptop PC, handheld device, tablet PC,touch screen or tone dialling system. Alternatively they can enter data indirectlyusing character recognition of manually transferred sheets of paper or faxes, orvoice recognition. All these methods result in the data being available for furtherprocessing without the need for further human intervention.

The value of EDC is only realised if it is integrated into further businessprocesses. This may mean workflow systems, post processing, automatic replysystems and organised and searchable data stores. These are described below:

Current UsageEDC is key wherever immediate access to live data collection is important. Thiscovers a wide variety of situations, in many industries. The following examplesseek to show just how widespread this is.

In the commercial world EDC is already widely adopted. When you are nextdoing your supermarket shopping, note how many of the assistants check whichitems are required to restock the shelves using handheld devices with built in barcode readers. Also, delivery drivers can now record the exact load delivered at asite rather than just what was ordered. This is automatically transferred back tobase, allowing the amended invoice to be printed and in the post (or sent via e-mail) before the driver has left site. In other areas, consumer feedback is nowgathered via web sites with questionnaires rather than phone conversations orpost-back forms.

In manufacturing, increasing amounts of data are gathered electronically. Manyassembly lines are dotted with touch screen devices to allow operators to inputperformance information for immediate analysis and audit trail generation.



Inspectors use hand held devices to record their findings as they travel aroundthe plant without the need to transcribe paper notes when they return to theiroffices.

The scientific world is starting to realise the benefits of EDC more fully. In thelaboratory, Electronic Lab Notebooks are becoming increasingly popular, drivenby improved data collection devices such as tablet PCs and better integration withlab instruments. In the field, data is now gathered using rugged hand held deviceswith bespoke applications, for example, recording habitat or land use information.

Perhaps the area where the most dramatic growth in EDC is set to occur isclinical trials data collection. Here the combination of tight timescales, reducedcosts and more accurate analysis of results is driving the investment required toimplement EDC in both large and small clinical trials over all phases of drugevaluation. Data collection can utilise many of the methods described above, withtone phone and fax based systems being overtaken by bedside hand held devicesand even direct data recording from sensors worn by the subject.

Key IssuesBefore discussing the possible solutions to the challenges of EDC, it is importantto consider the major issues. Some of the questions that need to be addressed arelisted below. As you can see this is a considerable list and serves to encourage afull review of the situation before embarking on an investigation of EDC.

� What data are you gathering? - Without a clear understanding of the datacharacteristics you are gathering you should not even start. Is the datastructured e.g. a questionnaire - or is it driven by outside activities, e.g. a labnotebook? Is it simple text, dates, numbers or more complex data such as barcodes, drawings, sound, photos or externally acquired data from a measuringdevice? What is the data to be used for?

� User Characteristics - The characteristics of the user population need to beconsidered carefully when investigating EDC. Are the users regular userswho will build up skills in the interface or are they one off users who need asimple interface? Are they computer literate or technically unskilled orresistant? What hardware do they have available to them, or will they neednew, specific hardware?

� Portability - Does the data need gathering in a particular location, e.g. outside,shop floor, hospital bedside, patient home? Does the user need to carry the



data-recording device with them or can they enter the data at their desks?� Connectivity - Can permanent network attachment be deployed, e.g. wire in

the wall or WiFi and indeed is it required? Are off-line connectionsappropriate (dial-up links, docking stations) and if so how are they achievedand how often? Is a public connection allowed (e.g. via the Internet) or mustthe data be transferred over secure networks? Is two-way communicationrequired or can all communication be initiated from the client end? What isstored locally and what is stored centrally?

� External devices - Does the user need to interact with external devices suchas bar code readers or special laboratory equipment? Is this automatically partof the workflow, e.g. reading a bar code then typing in a number?

� Validation - Is validation of input data required? If so how complex is it? Is itlimited to simple rules such as field1 must be in the range 1-100, or complexinterdependencies (field1 < field2 > field3*0.4)? When is validationperformed- immediately a number is entered or at a later time?

� Feedback - Is data entered one way (from the client end to the server) or isfeedback required, e.g. to instruct the user to perform an action based on theinput?

� Workflow - Is data collection the start of a complex series of steps arrangedinto a workflow or is it a one off action? Do the data need validating,reviewing, approving and releasing for general consumption locally andcentrally? Can further actions be automatically initiated based on the datacollection or any other action in the chain?

� Integration - Does the system need integrating with other computer systems?These may be databases, processing systems or additional data collectionsystems, e.g. the results of an assay or blood test requested at data collectiontime.

� Post processing. What analysis is performed on the data collected? Is thisperformed in real time or manually at a later date? What actions result fromthe processing, e.g. approving or rejecting a drug in a clinical trial?

� Regulations. Does the system have to conform to any externally imposedregulations, such as 21 CFR Part 11, Good Manufacturing Practice, or GoodLaboratory Practice? This can have a decisive effect on the approach to thesolution.

Possible SolutionsHaving analysed the problem, it is now possible to review possible solutions forthe use of EDC in a particular situation. Each approach has advantages and



disadvantages and indeed, may be used in parallel within the same overall solution.Current innovations, especially in the area of mobile computing, are making thechoices wider and the benefits greater. A good architecture removes the storage,processing and workflow decision-making from the front end so that differentfront-end devices can be swapped without changing the overall structure.

1. PC Client Applications

The easiest method of EDC uses traditional computer based forms to gather data.At its simplest this can be a set of Windows-based forms that lead the userthrough a questionnaire style interface. This approach, often described as Wizard,has the advantage that the user interface is function rich, easy to interact withand can perform a lot of cross checking and validation. PCs are widely available,laptops can be fairly portable and the user interface is familiar to most users. PCsare also able to connect easily to external devices such as bar code readers,scales, thermometers or even more complex scientific devices.

The problem with this approach is the need for an application to be deployed oneach PC. This may require a complex set-up or may have specific pre-requisites(e.g. operating system version) that may not be under central control. This canlead to complex system management challenges. Also, although a laptop isportable, it cannot be carried around when gathering data. PCs are genericdevices and so are not specifically set up for the job in hand and if each userrequires an extra PC they are expensive.

Despite these concerns, this approach is used by many organisations. Novartis,for example, use a clinical trials system based on PhOSCo (www.phosco.org),which is deployed on each investigator’s PC. As with many of the solutionsdiscussed below, connectivity to the local or central data stores and the ability towork offline are issues. In the PC context, significant amounts of data can bestored locally before transfer. This means the data collector can work off-linebefore uploading the data to a server at a convenient time.

2. Web Based FormsThe Internet offers great possibilities for allowing users to enter data directly intothe system via browser-based forms. If deployed over the web these are globallyavailable and users have good familiarity with the technology. No software isrequired in the client device (except a browser) making system management easyand reducing the need for expensive specialist hardware. A certain amount of



validation can be built in and data is available immediately on the server.Responses based on the input, for example, instructions on the next step, can begiven immediately.

Internet systems, however, have several drawbacks. The user must always beconnected to the network either by phone or LAN connection. This is not alwayspossible in hostile or remote environments. The system is dependent on theperformance of the network, which may result in slow responses. The userinterface is not as rich or elegant as a client based system, which can make datacollection slower and more awkward. If using the public Internet, data security isa major concern and needs special attention.

Despite these drawbacks, web based systems have a major part to play in EDC.They are especially useful where occasional users will not be able to installspecialist hardware or software. Even if not used in primary data collection, webbased interfaces are very useful for certain parts of the workflow such asreporting, graphing and review and approval cycles. Here managers tend to bepermanently connected to a network and there is a reduced need for rapid dataentry. Live access to the data collection realises the benefits of collecting datarapidly in the first place.

3. Touch ScreensTouch screens allow users to input data rapidly using bespoke forms. They caneasily be incorporated into bespoke hardware systems such as scientificinstruments or point of sale devices. Little or no computer literacy is required tooperate them and they can be deployed in places where more generic inputdevices such as keyboards or mice would be hard to use, for example, a factoryfloor or a public place. System management is easier as the users do not haveaccess to the system configuration data. Data can be sent directly to the serverfor analysis and action and responses sent immediately to the user if required.

Touch screen devices always require specialist hardware. Despite advances intouch screen technology, complex data entry is still error prone. They are onlyreally useful when the user has a limited number of choices to make and must beused in a very prescriptive manner.

Touch screen systems are best used in specific environments where explicit datacollection is needed. They are especially useful where the operators do not have



the time or the inclination to sit down and learn complex interfaces and do nothave the opportunity to spend a lot of time entering data.

4. Bespoke Handheld DevicesThe benefits of handheld devices are significant. They can be carried to the pointof data collection so errors are not introduced between recording the data andentering it onto the system. Examples are bedside patient data entry or takingreadings from industrial devices running in a production plant.

Bespoke devices are built for one purpose. The advantages of this are that theydo not contain unnecessary functions and are tuned for the needs of the user. Theinterface can be set up exactly for the job required and is often greatly simplified.They can be linked to other input devices, such as bar code readers, thusminimising the chance of incorrect data entry. They can be kept smaller, due tothe elimination of unnecessary functions, and can be built to specific standards,for example, for rugged environments.

These advantages are accompanied by certain drawbacks. Because they arespecific, generalising their use is difficult or impossible reducing flexibility or theopportunity for re-use. Most importantly, they do not benefit from economies ofscale and so purchase price is likely to be high unless ordered in bulk.

Bespoke devices have specific uses that will continue despite the growth ofcommodity handheld devices. However, they are likely to remain niche items ifcurrent trends continue.

5. Commodity Handheld Devices (PDAs)Recent years have seen a spectacular growth in the sales of commodity handhelddevices such as the Palm Pilot or the Compaq iPaq. Unit prices for these devices,commonly referred to as Personal Digital Assistants (PDAs), have fallensignificantly and powerful new features added, such as colour screens and morememory. The number of operating systems is falling and user acceptance is high.PDAs offer a much cheaper alternative to using laptop PCs in many EDCsituations.

EDC using commodity PDAs is a fast growth area. High quality interfaces can beproduced with built-in validation. Local processing and storage is possible andextra functionality can be added using extra cards to add, for example,



networking or mobile phone capabilities. The devices are truly portable and caneasily fit inside the pocket of a suit, overalls or lab coat. However, this portabilityinevitably means that the display area is very limited, which can be a significantlimitation to the development of good user interfaces.

Producing good EDC solutions using off the shelf hardware has its challenges.The operating systems are new and can often have problems, especially withconnectivity. Both software and hardware are changing rapidly and can constitutea moving target when developing software. If inappropriately designed, the userinterface can be fiddly to use leading to input errors, slow data entry or poor useracceptance. The need for extra cards to add functions can quickly make thesystems bulky. Battery life is good, but still not infinite and forgetting to put thedevice on to charge can cause problems.

Despite the challenges, commodity PDAs represent a major growth area for EDC.These systems reduce the investment barrier to using truly portable EDC and willmake a big difference to this area over coming years.

6. Mobile PhonesThe continual blurring between mobile phones and PDAs again offers possibilitiesfor EDC on the move. High-end phones now have processing capabilitiesallowing programs to be uploaded and run on them. They can be smaller andmore portable than PDAs and are, of course, fully connected. The growth offaster data connection routes such as GPRS and 3G means data upload can bevery quick. Photo capture is increasingly common in the more expensive phones.All phones have built-in messaging using SMS. User acceptance is high as fewpeople need to be persuaded to carry an expensive mobile phone.



Using phones for EDC is only appropriate in certain situations. The userinterfaces are very small and fiddly and not appropriate for collecting largeamounts of data. The connectivity is not always reliable and may not be permittedin certain environments such as hospitals or on aeroplanes. Developing bespokeapplications for phones is a new area and again may not be reliable.

In situations where users are required to take readings at specified intervals or tocarry out some other action (e.g. take a drug) they can easily be prompted to doso by phoning them up.

Mobile phones are likely to become more prevalent in EDC as their power grows.For situations where remote data entry is required and low functionality isacceptable, they will offer an interesting alternative to PDAs. Devices that can beconsidered both PDAs and phones, such as the O2 xda, will offer majorpossibilities in this area.

7. Tablet PCsThe recent release of Tablet PCs offers huge opportunities for EDC. Thesedevices are an interesting mix of laptop PC and PDA that present true portability,an intuitive interface and reasonably powerful processing. Still in their infancy,they are backed by many big players, most notably Microsoft. They offer goodconnectivity using wireless networking, fast data entry using handwritingrecognition and the ability to produce slick systems with good user acceptance.

The drawbacks of Tablet PCs may come from their newness. The marketplace isfairly volatile and likely to change rapidly. The handwriting recognition featuresmay only be useful in some circumstances (for example, free text and diagrams)and more traditional touch screen interfaces used for the bulk of data entry.Developers will have to gain experience at producing optimal user interfaces andusers will have to get used to using them.

The size of a Tablet PC, whilst smaller than a laptop, is considerably more bulkythan a PDA. However, the increased display area gives much better opportunitiesfor developing effective user interfaces. They are, after all, much more akin tothe familiar clipboard and paper pad. The connectivity benefits can only berealised in a networked environment.

The use of Tablet PCs for EDC is likely to evolve considerably over coming



years. In many circumstances where fast, complex data entry is required in aportable environment they will prove invaluable.

8. Paper Based MethodsWhilst considering many of the new and exciting methods discussed above,much EDC is still done using more traditional methods. Paper methods rely onforms being filled in by hand and then scanned into a computer, either locally ormore usefully via fax. When using the latter, replies can automatically be returnedby fax. The system must interpret the paper form and extract data from it,validating at this time. Although very much ‘old technology’ it is, however, veryportable, universally accessible and has no requirement for extra local hardwareor software (every ward in a global clinical trial may not have a PC, but it willalmost certainly have access to a fax machine). Access is full time (24/7) anddata entry is immediate at the centre. The original data submissions (fax images)can be retained for audit and regulatory purposes.

Paper based methods still have considerable limitations. Form design is critical toreduce data entry errors and may result in cumbersome forms. There is no datavalidation until data is submitted and the reply may take several minutes to beresent. Connectivity is error prone as return fax numbers are often wrong or dialout only.

Paper based systems, especially fax, may seem old technology but they can be aconsiderable improvement over existing off-line systems (manual collection andentry of paper based forms).

9. Electronic DocumentsOne alternative to paper is provided by allowing users to download electronicdocuments that are completed using the macros embedded within them, thenemailed or uploaded to the server. Current systems have been produced in AdobeAcrobat, MS Word, and MS Excel. This has the advantage that they can be filledin locally, require no additional local software and look like traditional paper basedforms. They are also very easily distributed.

The down side is the lack of advanced validation and the possibility the user willnot send the forms back. Also, some systems are insecure and allow incorrectdata entry, for example, into incorrect cells in the spreadsheet. The documentsmay require a specific version of the editing software (for example, Office XP



rather than Office 2000), which reduces their effectiveness.

Electronic document questionnaires are becoming more popular with theincreasing sophistication of macro languages. Given the widespread availability ofthe client software (for example, MS Office) this is likely to continue.

10. Sound Based SystemsAs with paper-based systems, telephones are a universally available deviceavailable at no extra charge. Entering data using either touch phone techniques orvoice recognition is available to all and can have built in validation. This method isalso very easy to use and familiar to just about everyone.

Phone based systems have their problems. Tone phone systems are very slowand frustrating to enter data with and can be error prone. Voice recognition isquicker but more error prone, especially where global data collection is requiredand users may have different accents or even not speak the base language.

Phone systems are still extensively used for EDC due to their universalaccessibility. It remains to be seen whether the newer forms of EDC start toerode their popularity.

Further Considerations

User CharacteristicsThe choice and design of the solution must take account of the type of personentering the data. Good technical solutions are of little benefit if the users cannotuse the system or use it incorrectly. This is especially important where use isvoluntary – hard to use systems will not be used at all.

Occasional users will not have time to learn complex input forms. They requiresimple forms with simple options. A series of questions, one per screen, is a goodway to get the data required. This is especially important when gathering datafrom technically inexperienced or resistant groups, for example, elderly patients.This becomes more important where the users are based away from localsupport, for example, at home or on the road.



Regular users will require efficiency when entering data. This is vital where datacollection is a large part of their daily work and may be seen as a burden if ittakes too long. This is especially important where users are familiar with paper-based systems (laboratory notebooks, manual patient records), which are quickand flexible.

Whatever the target user group, interface prototyping and investigation is vital



when designing these systems. The best judges of the usefulness of the interfaceare the people who will use it.

Form DesignForm design in EDC solutions is very important however the system isimplemented. This can be critical where poor form design skews the results andinvalidates the data. The different format of the data collection medium comparedto traditional paper based methods presents both designers and users withchallenges – smaller screens, different forms of data input and user resistance.Trying different approaches with real users and getting external validation shouldboth be investigated before agreeing a particular solution.

Any system designed will almost certainly need updating at some time. This isespecially true where one EDC exercise is only required for a small number ofcases, but that many such exercises are likely, for example, in Phase 1 ClinicalTrials. This makes it important to be able to alter the system to design new formswithout the need for system changes or costly third party support.

Design systems should allow you to maintain libraries of frequently used forms(for example, vital signs). They should allow you to apply validation rules, screenlayout, help systems and workflow. They should also allow you automaticdeployment to user platforms without the need for a complex rollout mechanism.

An example of a form designer used to produce the following user interface



Without an effective form customiser, companies can easily find themselveslocked into an expensive and unsatisfactory relationship with a supplier. However,such flexibility comes at a cost that can only be justified if required.

Input ValidationOne of the major benefits of EDC is that data entry can be validated immediatelyrather than some time later, when it is too late to re-collect the data. Most of thesystems described above have some form of validation. Individual fields can bevalidated against fixed or flexible boundaries or permitted dictionaries of values.They can also be validated against other fields, for example patient weight mustbe no more than +/- 10% of the previous occasion it was recorded.

Validation can also occur at form level or at entire data series level, for example,printing a warning message if the value is more than 2 standard deviations fromthe mean of previously collected values.

Workflow – Immediate MessagingOften when data is entered into the system, the user requires instructions as towhat to do next. This can be based on some simple logic using the input values oron a more complex analysis of the data. An example of the latter is active learningused to steer clinical trials. Depending on the input data, the user can be advised



on which dose of the trial compound to give the patient in order to gain mostinformation from the treatment. This can reduce trial time by up to 30%. Thissort of facility requires complex analysis but is only possible with live data entry.

Immediate messaging can also be used to activate support if there is a problem oropportunity. Problems can be trapped and dealt with immediately if an alert is sentto support personnel. Also, it is possible to track what is going on and discussthis immediately with the user which presents a real opportunity to, for example,encourage further use of the system.

Immediate messaging is only possible if the interaction device is attached to thedecision-making facility. This may be performed on a central server accessed, forexample, by fax, Internet, on a local off-line laptop or Tablet PC. Howeverperformed, the decision-making engine must be able to respond back to the userimmediately.

Workflow – Subsequent ActivitiesAs data is collected live it can be passed on to other parts of the organisationimmediately. Collection of shelf contents in a supermarket is useful for stafftasked with filling the stores, but it can also feed directly into the central orderingfacility that automatically requests replacements. Data collected in a laboratorymay need approval before it is released for general consumption.

Integration into further workflows is only useful if data is transferredautomatically and immediately to the next point in the chain. If data must betransferred manually (for example, manually attaching a document to an e-mail) itmay either not be done or may be delayed. Workflows should also be two way,allowing feedback to the user of the results of subsequent activities. Because thedata is collected so much quicker, decisions can be made using it without waitingfor it to be passed back.

Off-line vs. On-Line Data CollectionOne of the key driving forces behind decisions on technology is the connectivityof the data collection device. On-line systems such as client-server PCs, WiFinetworked PDAs, web browser systems or fax-based systems are always incontact with the server. Data is never saved locally, it is always sent through tothe server. The data is immediately available for further processing and responsescan be based on all other collected data. Messages can be sent back to the user



immediately (for example, via on-screen messages or faxes sent to the user).These types of systems are always preferable, but limitations may make their useimpossible.

Off-line systems first gather data locally then send it in to the central server at alater date. These may be required where continuous connection is impossible,such as in hospitals and in remote outdoor locations. Off-line systems allow thedata to be reviewed prior to dispatch. Depending on the power of the datacollection system, they can allow a great deal of validation and decision makinglocally. Data transfer is, however, delayed and may not occur if it is forgotten.Also, complex logic may be required to ensure the local and central systems arein step if both are permitted to edit the same data. Local data storage is requiredwhich may be quite large and could, for example, easily be beyond the capabilitiesof a PDA

Post ProcessingBecause the data is available centrally much more quickly than using traditionalmethods, post processing can be performed at a much earlier stage. This mayinclude statistical analysis or decision-making tools, or confirmation that therecorded data is within some specific targets. When adopting EDC it is importantto consider how to benefit from the timely gathering of the data to makedecisions earlier.

Integration with other Data Collection ProcessesWhilst EDC is an important part of the data gathering process, it is not the wholestory. Additional data may come from corporate databases, external data feeds orhardware devices that feed into the system. For example, consider a sample ofmaterial being submitted for assay. This fact is entered onto the system alongwith the barcode of the packet containing the sample. The sample is thendispatched to the lab for specific analysis. The results must be fed into thesystem and matched up to the original request. Full integration of the EDC,workflow and LIMS systems in this way can significantly reduce turnaroundtime and errors.

When planning fully integrated systems of this nature, it is important to allow forthe very different forms of data gathered in each cycle. Devices such as massspectrometers can gather large amounts of data that may not fit in with thesimple data structures of an EDC system without further processing. Also, lab



data may need to be specially integrated with the data structures of a fixed EDCsystem.

RegulationMany EDC solutions must be deployed in environments regulated by governmentagencies. These agencies have a good record for ensuring that companiesproduce secure and reliable systems, but the rapid growth in electronic systemsis making it difficult to keep up. Widespread use of technologies, such aselectronic signatures, is not advanced and thus proving an EDC system can bedifficult.

The US Food and Drink Administration (FDA) cast a long shadow overcompanies wishing to exploit EDC in the pharmaceutical sector. Although theaccompanying 21CFR Part 11 regulation appears to be in the process ofsimplification to stimulate development in this area, many organisations are beingheld back by the FDA’s unclear stance. This is an area that needs consideration atthe start of any investigation of EDC but should not be allowed to impede it.

Benefits of Electronic Data CaptureThe descriptions above demonstrate the wide benefits of EDC. The following listis a summary to help show how an investment in an EDC solution can quickly berepaid.

� Remove slow human input and data transfer stage - The manual re-entry ofdata recorded on paper is expensive and unreliable. EDC removes this stepaltogether.

� Immediate validation of input - Immediate validation of data entry andtransfer trap errors early and help to eliminate them.

� Data can be gathered in an easily processed form - Data entry is quick andefficient and tuneable to the user population.

� Instant access by co-workers - As soon as the data is on the system, otherstaff in the organisation can see it and use it.

� Data is searchable and easily categorised - This is especially important whencompared to paper records such as lab notebooks.

� Feed into workflow - Subsequent processing can proceed immediately thedata is received.

� Quick decision making - A critical benefit is that a decision point in a projectcan be reached much more quickly, saving both time and money.



Case StudiesThe following demonstrate how EDC has been used in practical situations andhow this has been a major benefit to organisations. The following shortdescriptions show ways in which it is being deployed and some of the benefits.

Clinical Trials using Adaptive LearningTo ensure they are statistically valid, clinical trials need large numbers of subjects(patients). This is especially true of Phase 2 (Dose calculation) and Phase 3(Effectiveness) trials. Advanced statistical techniques can be used to steer the trialto ensure the maximum information is extracted from the minimum number ofsubjects. The software allocates a dose of target compound or placebo based onthe patient condition and the gaps in the system’s knowledge.

To ensure this happens effectively, the information must be fed into a centralsystem where the calculation occurs. The response must be then sent back to theinvestigator. For conditions where immediate treatment is vital this must occurvery quickly, potentially from anywhere in the world. The results of the treatmentmust also be fed back quickly to ensure the knowledge base is kept up to date forfurther treatment allocations.

A major pharmaceutical company approached Tessella to provide a system toexploit these techniques. A phase 2 trial was to be deployed globally in around 200centres. The equipment in these centres was highly variable, and the lowestcommon denominator was a phone and fax system. The decision was thus takento use fax based data submission and dose allocation responses.

Tessella developed the fax handling software using an off the shelf OpticalCharacter Recognition (OCR) package that decoded the incoming faxes and sentresponses back again. We also developed a flexible database and a managementreporting and administration suite. The statistical calculation engine was providedby Duke University in the USA. The entire system was supported 24 hours perday by a Tessella team.

The clinical trial ran for 18 months, gathering data from approximately1000patients. Using the advanced analysis, 300 fewer patients were requiredbefore the end conditions were reached. The EDC approach also meant that thelast patient form was received only a few weeks after the end of the trial allowingfurther investigation decisions to be brought forward significantly.



The use of EDC had other benefits beyond the adoption of advanced statistics.The trial manager was able to maintain a much better picture of the progress ofthe trial and control progress accordingly. Also, he was able to react to eventsmore quickly, for example, phoning investigators who had submitted subjects, toencourage them to submit more. This and the benefits above led to a saving ofover $3 million.

This technique is now being generalised for other trials. Tone phone and PDAfront ends are being prototyped, and in some cases, deployed.

Macaulay Land Use Research InstituteThe Macaulay Land Use Research Institute (MLURI) is a multidisciplinary-basedresearch institute located in Aberdeen, Scotland. They carry out research into thephysical, environmental and social consequences of land use. Their researchstation in Hartwood, Lanarkshire carries out research on the grazing ecology ofsown swards and the development of sustainable management systems. Thisinvolves the collection of botanical information from a field.

Originally, MLURI’s Point software was used to record botanical information.The program was written in Pascal and had been run for many years on HuskyHunter handheld devices. The Hunters have now been replaced with Husky Fex21machines, which are handheld PCs running Pocket PC.

Tessella was asked to migrate the Point application to the new platform, makinguse of the Pocket PC Graphical User Interface (GUI).

Tessella developed Point2 for the new Husky Fex21 devices using MicrosofteMbedded Visual Basic, part of the eMbedded Visual Tools suite which isdesigned specifically for creating applications for Pocket PC devices.The new Point2 application for the Husky is a significant improvement over theold application that ran on the Hunter machines. Point2 stores the field data as atext file, which can be easily transferred to a desktop computer for furtheranalysis. The text files can be saved to the device’s ROM or to a storage card ifinstalled.

The provision of a GUI makes the application much easier and quicker to use.Now, instead of having to enter data into the system via the keyboard, MLURIresearchers can use the stylus to interact with the various elements of the GUI.This stylus is the portable version of the mouse and facilitates quick data entry,no matter how unfavourable the physical environment.



Unlike the desktop PC, portable computers can vary significantly in terms ofprocessor platform, memory capacity, and input/output strategy. As a result,designing applications to run on a spread of vendors’ hardware can bechallenging. This is partly addressed by the eMbedded Visual Tools, which canprovide a framework for developing for a selected target platform.

Clinical Trials(Source: Bio-IT World)Sylvia Collins, Novartis Vice president has implemented EDC clinical trialsolutions at two major pharmaceutical companies. Whilst at Bayer, she worked onthe adoption of EDC solutions and by 1995 100% of trials used EDC. Timebetween the last patient visit and database lockdown was reduced from anaverage of 12 weeks for paper-based systems to just 4 weeks. Now at Novartis,Collins says that the lag time has reduced yet further and she can anticipate a timewhen it is reduced to just 1 day.

Novartis has taken the important step of bringing the ownership of the product inhouse. Based on the popular PhOSCo product, the system is continually beingdeveloped. It is deployed on 3000 laptops worldwide and had been deployed on150 trials within 18 months. Database licence savings alone have been over $50million. The configuration and deployment of the clinical report forms takesaround 9 weeks but Novartis hopes to improve this considerably.

Implementing a system of this scale was a major challenge, according to Collins.Strong leadership, change management and broad authority were required. Mostimportantly it needs to be pushed forward as a business development, not an ITdevelopment.

The PhOSCo product is available for purchase, on a per trial or corporate basis,and is supplied with full Java source code for customisation and integration.

Mobile Data Entry and InvoicingBottled water delivery companies often incentivise their delivery drivers byoffering bonuses based on persuading customer to take more than they ordered.This is very successful but creates a paperwork headache as the invoice cannotbe printed and posted until the driver returns to base and indicates how manybottles and cups were actually delivered, rather than the number ordered.



One of Tessella’s customers, a world leader in bottled water delivery, isdeveloping a system for providing drivers with PDAs connected via mobile phonecards to base. Once a delivery has taken place, the driver enters the amountactually sold onto their PDA. This is sent back to base using the GPRS card andis passed into the Oracle Financials database. The invoice can then be printed andposted before the driver is back at the depot, improving cash flow and reducingcredit times significantly.

The system has other benefits. Routing information can be passed to the PDA sothe driver has the system in the vehicle. Also, emergency delivery informationsuch as extra deliveries or cancellations can be sent to the cab, allowing for extrasales or removing wasted journeys.

What the future holds for EDCElectronic Data Capture is set for rapid growth over the next few years. Forexample, EDC use in clinical trials has moved from 12% to 24% in just two years.Many current developments are expected to fuel the growth and will enablecompanies to achieve large returns on investment in a short timescale.

A leading stimulation to growth will be the reduction in price and increasedsophistication and power of small handheld devices. This will include laptop PCs,PDAs and especially tablet PCs, which are expected to make a significantcontribution in this area.

An aspect of technology that will also drive growth in EDC is improvements inconnectivity. Increased use of GPRS and 3G mobile connections and wideraccess to WiFi will make communicating with central resources especially easy.Also, more and more devices will have connectivity built in with wireless cardsdelivered with the system.

These developments are only useful if accepted by the users. A general wideracceptance of technology and especially hand held devices will make take upeasier and acceptance more straightforward.

Increased integration with other forms of data collection will be lead by thewidespread adoption of RF Tags. This will enable the easy combination of humanand automated data collection in a simple, single system.



Software improvements will also continue. User interface design will improve,but so will customisation and form design, and integration with other businesssystems as EDC becomes a mainstream part of the business rather than aninteresting add-on.

� Archiving of Electronic Info� Active Server Pages� Automated GUI Testing� Bayesian Statistics� Beowulf Clusters� C++� Client-Server Technology� COM� Computational Fluid Dynamics� Computer Image Processing� Decision Support Systems� Electronic Data Capture� Electronic Lab Notebooks� Excel� Extending the Life of Software� Federal Drug Administration� FORTRAN 90� Grid Computing� High Throughput Screening� Instrumentation� Integrated Lab Systems� J2EE� Java� Lims� Linux� Microsoft Net

� Object Oriented Programming� Pocket PC� Portable GUI Development� Printer Technology Guide� Real Time Systems� Regression Testing� Security and the Internet� Simulation� Soft Computing� Software Design Methodologies� Software Development Cycle� Software Documentation� Software Portability� Software Re-engineering� Software Specification� SQL� UNIX Inter-Process Comms� UNIX Systems Performance� UNIX Workstations� Visual Basic 6� WAP� Web Services� Windows 2000 Services� XML� X Windows

Tessella Support Services plcCreating Software for Science and Engineering

Tessella’s services range from feasibility studies, through system design,development, implementation and ongoing support. Our expertise includes:

Data Analysis SoftwareData Capture

Simulation SoftwareAdvanced Graphics

Systems SupportDatabase Applications

Other Technical Supplements available include:

Tessella Support Services plc3 Vineyard Chambers, Abingdon, Oxon, OX14 3PX, England

Tel: (+44) (0) 1235 555511 Fax: (+44) (0) 1235 553301E-mail: [email protected] Web Address: http://www.tessella.com